Shear stress transfer across concrete-to-concrete interfaces: Experimental evidence and available strength models
Otgonchimeg Davaadorj, Paolo M. Calvi, and John F. Stanton
■ This paper presents a study of 509 interface shear friction test results collected from literature on previ-ous interface shear experiments conducted between the 1960s and 2017.
■ The data collected were analyzed to gauge the reliability of six international code provisions and 14 strength models.
■ The analysis results identified the best-performing models for different scenarios, critical knowledge gaps and future research needs, and recommen-dations for ways current models could be further improved to achieve higher performance.
In structural concrete, shear force must sometimes be transferred across an interface between two materials. The interface may be between two faces of a crack in
monolithic concrete, two concretes cast at different times, or steel and concrete. Such shear transfer is usually modeled as a shear friction phenomenon. This approach, initially proposed in the 1960s by Birkeland and Birkeland,1 states that the shear strength of a concrete-to-concrete interface comes from the contribution of several resisting mechanisms, namely the co-hesion between particles, the friction between concrete parts, and the shear force resisted by the reinforcement crossing the interface. The empirical parameters involved have been calibrated against experimental evidence by numerous inves-tigators.1–29 Today, the shear friction theory is widely accepted and has been adopted by most design codes, including the PCI Design Handbook: Precast and Prestressed Concrete,30 the American Association of State Highway and Transporta-tion Officials’ AASHTO LRFD Bridge Design Specifications,31 and Building Code Requirements for Structural Concrete (ACI 318-14) and Commentary (ACI 318R-14).32
However, the various models differ in the ways they account for the controlling parameters and, in some cases, they provide significantly different strength estimates for the same configuration. For instance, ACI 318 advocates the use of a pure friction approach, the AASHTO LRFD specifi-cations recommend a cohesive component and a frictional component, and the PCI Design Handbook uses an effective friction model in which the shear strength is nonlinearly re-lated to the amount of reinforcement crossing the interface.
PCI Journal (ISSN 0887-9672) V. 65, No. 4, July–August 2020.
PCI Journal is published bimonthly by the Precast/Prestressed Concrete Institute, 8770 W. Bryn Mawr Ave., Suite 1150, Chicago, IL 60631.
by authors of papers in PCI Journal. Original manuscripts and discussion on published papers are accepted on review in accordance with the
Precast/Prestressed Concrete Institute’s peer-review process. No payment is offered.
88 PCI Journal | July–August 2020
For low steel ratios, the PCI Design Handbook and AASHTO LRFD specifications methods tend to give shear strengths significantly higher than the ACI 318 pure friction approach. ACI 318 separately addresses the issues of interface shear transfer and shear friction: shear is transferred across an inter-face for both, but the latter contains reinforcement across the interface while the former does not. Some of these inconsis-tencies have been discussed by others (such as Tanner33), but at present, there is no general agreement as to which design or analysis approach consistently provides the most accurate predictions and should thus be used.
This paper presents and critically analyzes the results of 509 shear friction push-off and pull-off tests collected from the literature, many generated before recent code updates.
The database analysis used the data to gauge the importance of various structural parameters and to evaluate the perfor-mance of six international code provisions and 14 of the main strength models proposed by different authors. The analysis results identified the best-performing models for different sce-narios, critical knowledge gaps and future research needs, and recommendations for ways current models could be further improved to achieve higher performance.
Database definition
A survey of the literature showed that experimental data pertaining to concrete-to-concrete interface tests have been conducted using the six test configurations outlined in Fig. 1. The database presented in this section addresses only direct
Figure 1. Common test configurations. Note: L = beam length; P = applied load.
89PCI Journal | July–August 2020
push-off and pull-off tests (cases in the top row of Fig. 1), interfaces subject to monotonic pure shear loads, and steel re-inforcement normal to the interface. These choices were made in order to focus on the basic shear transfer phenomenon. The main variables considered were the following:
• interface type
• concrete strength and weight
• shear interface area
• aggregate size
• reinforcement ratio, reinforcement strength, and average clamping stress
Many studies have focused on the effects of other more spe-cialized variables on interface transfer, but they were deliber-ately excluded from this database because of being out of the scope previously mentioned. Examples include the following:
• presence of axial loads (tensile or compressive) acting normal to the shear interface (Mattock and Hawkins,9 Mattock et al.,11 Walraven and Reinhardt,14 Papanicolau and Triantafillou,21 and Echegaray et al.27)
• orientation of reinforcement crossing the interface (Vang-sirirungrang,6 Dulacska,8 Mattock and Hawkins,9 Haw-kins and Kuchma,23 Nagle and Kuchma,34 and Mattock35)
• simultaneous presence of shear forces and bending mo-ments (Mattock et al.11)
• use of different materials, such as ultra-high-performance concrete, on the two sides of the interface (Crane36)
• presence of sustained loading prior to loading to failure (Frenay37)
• application of cyclic loading (Frenay,37 Valluvan et al.,18 and Calvi et al.38,39)
The resulting database of shear tests involved a total of 509 monotonically loaded pull-off and push-off tests, summarized in Table 1. The information reported in Table 1 is limited to a chronological list of the studies considered with the variable ranges explored in each. More details on the individual tests are provided in the work of Davaadorj.40 Table 1 gives details of a subset of the test programs that satisfy the data selection criteria of this research work. The total number of tests con-ducted in each of the programs listed may be higher than what is reported in the table (that is, all of the studies are included, but not necessarily the total number of tests from each study).
The experimental programs in the database involved speci-mens made of three types of concrete and four interface types. The concrete types were normalweight, sand lightweight, and
all lightweight (referred to in Table 1 as NW, SLW, and ALW, respectively). The interfaces were monolithic uncracked (MO-U), monolithic precracked (MO-P), and cold joints that were intentionally roughened (CJ-R) and not roughened (that is, smooth) (CJ-S).
Additional notation used to identify all other variables in Ta-ble 1 included compressive strength of concrete f c , maximum aggregate size a
g, reinforcement ratio ρ, reinforcement yield
stress fy, and average clamping stress ρf
y.
For cold-joint specimens with different concrete strengths on the two sides of the interface, only the lower compressive strength was reported. No additional restrictions were adopted for all other parameters, hence all values listed in Table 1 are those found in the original publications.
Of the 509 tests included in the database, 354 were normal-weight concrete, 103 were sand lightweight concrete, and 52 were all lightweight concrete (70%, 20%, and 10% of the total, respectively).
Figure 2 summarizes the other key variables investigated ex-perimentally over the course of past studies and provides the distributions and cumulative curves of the number of tests as a function of interface types, maximum aggregate size, concrete compressive strength, reinforcement yield stress, reinforce-ment ratio, and clamping stress.
Additional important parameters considered but not outlined in Fig. 2 were the area of the shear interface ranging from 50 to 160 in.2 (32,258 to 103,226 mm2) and the size of the reinforcement, which did not vary over a wide range. In fact, the bar diameters presented in the database varied between 0.2 and 0.63 in. (5.08 and 16 mm), with the majority of the tests conducted on specimens with a diameter of steel reinforce-ment d
s ≤ 0.5 in. (12.7 mm).
The multiparametric distribution depicted in Fig. 2 outlines that a substantial number of tests have been conducted in all of the four major surface types. The majority of tests available (ap-proximately 60% of the total) comprised monolithic specimens, with the most conducted on precracked specimens. Cold-joint specimens represented approximately 40% of the total collect-ed, and the majority were intentionally roughened cold joints.
The concrete strength varied over a wide range, with about 70% of the specimens having a strength between 3000 and 7500 psi (20.7 and 51.7 MPa). Roughly 20% of the speci-mens tested had a compressive strength between 10,000 and 18,000 psi (68.9 and 124.1 MPa), while relatively few speci-mens had strengths between 7500 and 10,000 psi—about 10% of the total. Of all these, approximately 80% of the specimens were made of aggregates with a maximum size between 0.375 and 0.75 in. (9.525 and 19.05 mm).
In approximately 90% of the tests, the reinforcement yield stress was lower than 80 ksi (551.6 MPa).
90 PCI Journal | July–August 2020
Table 1. Summary of interface shear experiments database
Authors YearNumber of tests
Concrete type
Interface type
Bar size, in.
ρ x 10-3 fy, ksi ρfy, ksi , ksi ag, in.
Hanson 1960 38 NW CJ-R, CJ-S ½ 0 to 8.18 47.0 to 52.0 0 to 0.4 3.0 to 4.2 ¾
Hofbeck et al. 1969 31 NW MO-U, MO-P ⅜ 4 to 26.0 42.4 to 66.1 0.2 to 1.4 2.4 to 4.5 ⅞
Mattock and Hawkins
1972 8 NW MO-U, MO-P ⅜ 8 to 18.3 49.5 to 53.7 0.4 to 1.0 3.9 to 5.1 n/a
CTA 1974 2 NW CJ-S n.d. 0 n.d. 0 4.0 to 5.5 n/a
Mattock et al. 1975 4 NW MO-U, MO-P ⅜ 10.5 to 16 50.1 to 52.7 0.5 to 0.8 3.8 to 4.2 ¾
Mattock et al. 1976 66NW, ALW, SLW
MO-U, MO-P ⅜ 0 to 26.4 47.7 to 53.6 0 to 1.4 2.0 to 6.0 ⅜ to ¾
Figure 2. Frequency distribution of tests as a function of main parameters. Note: CJ-R = cold joint intentionally roughened; CJ-S = cold joint not roughened, or smooth; MO-P = monolithic precracked; MO-U = monolithic uncracked. 1 in. = 25.4 mm; 1 ksi = 6.895 MPa.
92 PCI Journal | July–August 2020
The tested reinforcement ratios ranged from 0% to 3%, with most of the tests falling in the 0% to 2% range. This, com-bined with the reinforcement strengths considered, resulted in clamping stresses rarely exceeding 1600 psi (11 MPa). The few tests that exceeded this threshold were for monolithic precracked interfaces.
As shown in Table 1, all experimental programs that con-stitute the database were carried out between the 1960s and 2017. Figure 3 and Table 2 provide a sense of how inter-face research has evolved over time and show that interface shear has consistently been a topic of interest over the past six decades, with substantial research carried out in the past 10 years for both normalweight and lightweight concrete (approximately 180 tests in total).
Although normalweight concrete has evidently received greater attention, notable studies on lightweight concrete have been conducted both in the 1970s and recently (for example, Sneed et al.28).
Alongside the distribution of shear friction tests over time, Fig. 3 provides a summary of the major milestones (listed in Table 2) pertaining to interface shear research, as well as to the applicable shear friction code provisions as they
have changed over the years. Figure 3 and Table 2 present the amount of data available at times when models and code provisions were proposed or changes were implemented. For example, the very first ACI 318 shear friction chapter was introduced when only 69 documented direct-shear transfer tests (normalweight concrete only) were available. As more studies became available, codes were updated accordingly, but some design limitations introduced decades ago are still enforced today.
Database analysis
The results of the database experiments are analyzed in this section as a function of the key parameters most likely to affect the interface behavior, with particular attention to the peak shear strength.
Figure 4 shows the experimental peak shear stress recorded for each specimen in the database as a function of the clamp-ing stress. Because all interface types were represented on the same graph, the results showed significant scatter.
In addition, the results were not separated as a function of any of the test variables. Trends observed were that concrete strength appeared to consistently increase with increasing
Figure 3. Distribution of shear friction tests in time. Note: LW = lightweight; NW = normalweight.
93PCI Journal | July–August 2020
clamping stress, and the raw data suggested that lightweight concrete specimens may be somewhat weaker than their nor-malweight concrete counterparts.
The same experimental results, separated as a function of the interface type, are shown in Fig. 4. Looking at the various interface types separately allowed a much more rational com-parison of the results. The trends were more apparent, but the scatter was still significant in all cases.
The amount of test data pertaining to monolithic interfaces (280 tests in total, with 89 uncracked tests and 191 precracked tests) surpassed that of cold-joint interfaces (229 tests, with 151 tests of intentionally roughened interfaces and 78 tests of interfaces that were not). However, enough test results were available for all interface types to allow the observation of the main trends. In particular, many tests have been conducted with clamping stress ρf
y values ranging from 0 to 1500 psi (0
to 10.3 MPa), with only a few outside this range. Two trends
Table 2. Shear friction research major milestones
Reference number in Fig. 3
Year Event description
1 1960 Hanson (1960) test program takes place.
2 1966First introduction of shear friction theory: Birkeland and Birkeland (1966) followed by Mast (1968) and Hofbeck et al. (1969).
3 1971 ACI 318’s first shear friction chapter (ACI 318-71).
4 1972Mattock and Hawkins (1972), followed by Mattock et al. (1975), study effects of external normal force.
5 1976 Mattock et al. (1976) first study influence of lightweight aggregate.
6 1978Shaikh’s (1978) effective shear friction concept is introduced, relying on the works of Mattock (1974), Birkeland and Birkeland (1966), and Raths (1977).
7 1978First appearance of effective shear friction in PCI Design Handbook, second edition. Shear fric-tion coefficients of cold joints that aren’t roughened and steel interfaces are 0.4 and 0.6, respec-tively. Corresponding upper limits are smaller than in current PCI Design Handbook.
8 1983ACI 318 adopts constant cohesion terms (based on aggregate type) in addition to the friction component. Minimum clamping force is set to 200 psi.
9 1985 Studies on effects of high-compressive-strength concrete begin.
10 1985PCI Design Handbook adopts higher upper limits. Shear friction coefficients for cold joints that aren’t roughened and steel interfaces are raised. Upper limits on effective shear friction coeffi-cients are introduced.
11 1994AASHTO LRFD specifications, first edition: Shear friction method that utilizes cohesion and external normal force. To take concrete/aggregate type into account, λ factor is used.
12 2008Tanner (2008) reviews PCI Design Handbook fourth, fifth, and sixth editions and addresses math-ematical flaws.
13 2008 ACI 318 adopts various shear strength upper limits based on interface type.
14 2010Significant lightweight concrete studies are conducted: Scott (2010), Sneed and Shaw (2014), Sneed et al. (2016), and others.
15 2011PCI’s effective shear method becomes applicable only to monolithic and intentionally roughened cold joint interfaces. λ2 term in equation of μeff becomes λ.
16 2014AASHTO LRFD specifications change to tabulated cohesion, µ, and upper limits for each inter-face and concrete type.
17 2014 ACI 318 equation form drops the cohesion term and becomes only dependent on friction term.
were detected from the whole data set: the interface strength appeared to increase with increasing clamping stress, and the lightweight concrete specimens appeared to fail at lower shear stresses than their normalweight counterparts. As discussed later, this may not have been due to the concrete type but rath-er other properties, such as the concrete strength. In addition, the specimens’ strength appeared to be affected by the inter-
face type, though at this stage it was difficult to quantify the extent to which that was true given that many other variables (such as concrete strength and steel strength) may have been the reason for a higher or lower observed response.
One important observation was that the experimental data for not roughened cold joint interfaces was somewhat sparse,
Figure 4. Interface shear strength as a function of clamping stress. Note: CJ-R = cold joint intentionally roughened; CJ-S = cold joint not roughened, or smooth; fy = yield stress of reinforcement; LW = lightweight; MO-P = monolithic precracked; MO-U = monolithic uncracked; NW = normalweight; vexp = shear strength measured experimentally (stress); ρ = reinforcement ratio; ρfy = clamping stress. 1 psi = 6.895 kPa.
95PCI Journal | July–August 2020
with normalweight data only associated with clamping stress-es less than 600 psi (4.1 MPa) and lightweight data mainly associated with clamping stresses greater than 600 psi. It was difficult to draw definitive conclusions regarding the influence of lightweight concrete on the shear strength of not roughened cold joint interfaces. The not roughened cold joint results were affected by significant scatter, with some of the normal-weight tests failing at surprisingly low stress. The problem was further exacerbated by the poor definition of a not rough-ened cold joint interface, namely, any cold joint that did not satisfy the 1⁄4 in. (6.35 mm) roughness requirement. This led to a variety of different surface conditions, all of which were treated as being in the same smooth joint category.
In contrast, the data pertaining to monolithic and intention-ally roughened cold joint specimens were extensive for both lightweight and normalweight concrete at all clamping stress values lower than 1500 psi (10.3 MPa). This allowed for a more thorough and rational analysis of the results, which will be discussed later. At this stage it was evident that, when broken out by concrete and interface type, the results shown
in Fig. 4 may be misleading, given that the strength of the var-ious specimens was affected by a number of additional hidden variables, such as concrete and steel strength. These aspects are discussed in the next two sections of the paper.
Effects of concrete strength
The analysis of the available results, as well as the conclu-sions set forth by the authors of previous studies, suggested that although the clamping stress is the main parameter affecting the shear strength of a given interface, the concrete compressive strength played an important role as well. The influence of this parameter on the shear strength of different interface types is examined in this section.
To simultaneously illustrate the influence of clamping stress and concrete strength, the experimental data presented two-di-mensionally in the previous subsection were examined in three dimensions. Figures 5 and 6 show the experimental shear strength of all four interface types as a function of f c and ρf
y.
Three-dimensional surfaces (left side of the figure) were also
presented to illustrate the main trends. These surfaces were based on extrapolation when no data points are available. They were intended to provide qualitative interpretations of the experimental results and may not necessarily provide accurate strength values for regions of the graphs with scarce experimental evidence. To provide a more complete overview of the results, the data were also shown in the form of contour plots (right side of the figure).
Figure 5 outlines the results pertaining to monolithic un-cracked and monolithic precracked interfaces, and the follow-ing major trends were observed:
• The shear strength vexp
generally increased with increas-ing ρf
y for both monolithic uncracked and precracked
specimens. In the context of shear friction, this trend indi-cated a positive friction coefficient.
• For monolithic uncracked specimens, the shear strength was not zero when ρf
y = 0, suggesting that some cohesive
component of shear strength existed. In contrast, while no monolithic precracked specimen tests with ρf
y = 0
were available, the data suggested that very small to no cohesion could be expected for monolithic precracked specimens. This was consistent with the notion that cohe-sion will not exist across an open crack.
• For monolithic uncracked specimens, the shear strength generally increased with increasing f c . Interestingly, the rate at which the specimens gained strength ap-peared to decrease at high values of ρf
y and f c . This
suggests that there may have been an absolute upper bound on shear strength, or an upper bound on the usable value of f c . However, the data were too sparse to be conclusive.
• The concrete strength f c did not appear to affect the shear strength of monolithic precracked specimens, par-ticularly for clamping stress values lower than 1500 psi (10.3 MPa). For higher clamping stresses, increases in
f c appeared to provide some benefit, but the tests in that region were few, trends were not clear, and only a handful of tests were available in that region of the graph.
• For both interface types, lightweight and normalweight specimens with similar f c values had approximately the same peak strength.
Figure 6 outlines the results pertaining to intentionally roughened and not roughened cold joint interfaces. Overall, the cold joint data were less voluminous than the mono-lithic data. The surface conditions for cold-joint interfaces were generally not well defined in test programs beyond the distinction of being intentionally roughened or not. Both joint features implied greater scatter and less reliability. This was particularly true for the data on cold joints that were not roughened, which were overall deemed insufficient to draw any reliable conclusions.
For cold joint interfaces, the following major trends were observed:
• The shear strength vexp
generally increased with increas-ing ρf
y for both interface types.
• For intentionally roughened cold joint specimens, the shear strength was not zero when ρf
y = 0. This was con-
sistent with what was observed for monolithic uncracked interfaces and suggested that some cohesive component of shear strength existed. However, the cohesion appeared to be lower than that observed in the monolithic un-cracked case and did not increase with increasing f c . In contrast, very low to no cohesion was observed for spec-imens that did not have roughened cold joints, with very scattered results, which was probably due to the inherent variability in the roughness conditions.
• Similar to what was observed for monolithic uncracked specimens, the shear strength tended to increase with increasing f c . The highest shear strength was recorded for specimens that simultaneously used high-strength concrete and high-strength steel reinforcement.
• No consistent trends were observed with respect to the in-fluence of the concrete strength f c on the shear strength of cold joints that were not roughened. Overall, data for cold joints that were not roughened were too sparse and scattered to allow any definitive conclusions to be drawn. Normalweight data were available only for clamping stress values lower than 600 psi (4.1 MPa), whereas lightweight data were only available for higher clamping stresses. In addition, the results for the few high-strength normalweight concrete specimens failed at surprisingly low shear-stress values. The scatter and inconsistency of the results was attributed to the variability of the interface roughness (simply referred to as “not intentionally rough-ened”), which may have been drastically different in the different studies conducted.
• For the monolithic interface types, the use of lightweight concrete did not seem to make a difference with respect to the peak strength for intentionally roughened cold joint interfaces. Again, the shear strength of lightweight specimens was consistent with that of normalweight specimens with similar concrete strength. Trends were not observed for specimens with cold joints that had not been roughened.
Effects of reinforcement yield strength fy
Reinforcement congestion is often a problem in reinforced concrete structures, particularly at joint locations, connec-tions between slabs and shear walls, and the like. Thus, a reduction in the amount of reinforcement necessary to ensure an effective transfer of forces between elements is highly desirable.
One way of achieving this is to use high-strength steel shear reinforcement in place of regular-strength bars. The use of high-strength reinforcement in interface shear-transfer application has received increasing research attention in recent years (for example, Barbosa et al.41). However, an upper limit on the yield strength permitted for calculating shear strength is still enforced in the applicable code pro-visions, and reinforcement with f
y > 60 ksi (413.7 MPa) is
still not permitted.
An analysis of the assembled database revealed that only a handful of tests that used reinforcement with f
y >100 ksi
(689.5 MPa) were performed in past studies, hence nothing conclusive can be said on the implications of using them in practice. The great majority of the available test data was from specimens with f
y < 80 ksi (551.6 MPa), while about 10% of
the specimens (roughly 50 in total) had steel reinforcement with yield strengths between 80 and 100 ksi.
To isolate the shear-strength contribution of the steel rein-forcement strength from that of the concrete strength (which was shown to play an important role, particularly pertaining to certain interface types), the results summarized in Fig. 7 are normalized with respect to the square root of the concrete compressive strength f c , typically associated with the con-crete tensile strength. This normalization process resulted in a much lower scatter for all interface types.
As shown in Fig. 7, for a given clamping stress, concrete strength, and interface condition, the reinforcement yield stress f
y did not seem to have any particular effect (beneficial
or detrimental) with respect to the interface shear capaci-ty. This suggests that using steel reinforcement with yield strength beyond 60 ksi (413.7 MPa) (and particularly Grade 80 steel [551.6 MPa]) should potentially be allowed. This observation was in agreement with recent findings of Barbosa et al.,41 who observed that an increase in f
y (Grade 80 versus
Grade 60) resulted in equal or slightly higher interface shear capacity. (The authors’ conclusions were limited to no. 4 [13M] and no. 5 [16M] reinforcing bars.)
98 PCI Journal | July–August 2020
Effects of other parameters
In line with the conclusions of previous research studies, the analysis of the database assembled in this research program revealed that the most critical parameters influencing the shear strength of an interface were the clamping stress, the concrete compressive strength, and the interface type. In con-trast with current shear friction code provisions but consistent
with recent research findings (such as Sneed et al.28), using lightweight concrete did not seem to have any effects on the interface strength.
A number of other parameters may influence the inter-face shear strength but were not considered in this study. Examples include the size of the reinforcement crossing the interface, orientation of the reinforcement, aggregate
size, loading protocol, and loading conditions.1–33 In many instances, the evidence collected was not sufficient to draw definitive conclusions. For instance, bar sizes ranging only from no. 3 to 5 (10M to 16M) were used in previous stud-ies, and this range was too small to identify clear trends. Analogously, while it is known that cyclic loads tend to lower the interface strength, the experimental programs that considered cyclic loading conditions as a variable were substantially fewer than the studies that looked at mono-tonic loads. As a result, there is still debate as to how the cyclic nature of the loads should be incorporated in rational interface behavior models.
Loads applied normal to the interface contribute to increase (if compressive) or decrease (if tensile) the interface strength, and inclined reinforcement is generally believed to strengthen
the interface in shear, though there is debate as to whether this conclusion holds true in the presence of reversed cyclic loads. Both phenomena are included in the equations provided by ACI 318.32
Further research is necessary to evaluate or confirm the effects of these and other parameters and to establish how they should be incorporated into a rational interface behavior model and simplified code-oriented strength models.
Performance of available strength models
In this section, the capabilities of a number of interface strength models are evaluated against the experimental results collected in the database described in the previous sections.
Table 3. Performance of tested design and strength models
The models considered include the provisions contained in six major international codes—namely, the PCI Design Hand-book (sixth and seventh editions),29,30 the AASHTO LRFD specifications,31 ACI 318,32 the Canadian Standards Associa-tion’s Design of Concrete Structures,42 CEB-FIP Model Code 1990,43 and Eurocode.44
A total of 14 additional strength models proposed by various authors between the 1960s and 2018 were also considered. (See the list reported in Table 3.)
For the sake of brevity, only the details pertaining to the three U.S. codes and design manuals are reported in this section. For details pertaining to all of the other models, the reader is invited to refer to the original publications or to the work of Davaadorj,40 which provides a comprehensive summary.
ACI 318-14
According to chapter 22.9 of ACI 318, the shear stress that can be transferred across an interface subjected to pure shear and crossed by perpendicular steel reinforcement should be calculated as
vn = ρf
y μ
where
ρ = interface reinforcement ratio
fy = reinforcement yield stress (limited in design to 60
ksi [413.7 MPa])
μ = interface friction coefficient given as a function of the interface type (as reported in Table 4) and the concrete weight reduction factor λ (equal to 1.0 for normalweight, 0.85 for sand lightweight, and 0.75 for lightweight concrete)
Table 4 also reports the various strength limits.
PCI Design Handbook, seventh edition
The seventh edition of the PCI Design Handbook30 provides two equations to compute the shear stress that can be trans-ferred across an interface subjected to pure shear and crossed by perpendicular steel reinforcement:
vn = ρf
y μ
e (1)
and
vn = ρf
y μ (2)
where μe in Eq. (1) is the effective coefficient of friction, and
is given by
µe =φ1000λµ
vuwhere
φ = strength reduction factor
vu = factored shear stress demand, psi
1000 = units of psi
Limiting values for μe are reported in Table 5 alongside the
μ values for the various interface conditions. Table 5 also reports the various strength limits.
The basic coefficient of friction μ already includes the light-weight concrete factor λ. This approach, in which the effec-tive coefficient of friction (a component of the capacity) is a function of the demand V
u, can be rearranged to give
vn = 1000ρ fyλµ
This makes the shear strength vn proportional to the ρ fy
rather than ρfy, which has the effect of increasing shear
strengths at low clamping stress values and therefore has some resemblance to the addition of a cohesion term.
Table 4. ACI 318 friction coefficients and upper limits
Case Interface type µ vn,max
1 Concrete to concrete, cast monolithically 1.4λ
For normalweight concrete (monolithic or
roughened): least of
0.2fc'
(480+0.08fc')
1600
⎧
⎨⎪⎪
⎩⎪⎪
⎫
⎬⎪⎪
⎭⎪⎪
For all other cases: lesser of0.2f
c'
800
⎧⎨⎪
⎩⎪
⎫⎬⎪
⎭⎪
2 Concrete to hardened concrete with roughened interface 1.0λ
3Concrete placed against hardened concrete that is not intentionally roughened
0.6λ
4Concrete anchored to as-rolled structural steel by headed studs or by reinforcing bars
Equation (1) is recommended for use with monolithically cast interfaces and intentionally roughened cold joints, while Eq. (2) should be used for nonroughened cold joints and steel-to-concrete interfaces.
The AASHTO LRFD specifications recommend that the shear stress that can be transferred across an interface subjected to pure shear and crossed by perpendicular steel reinforcement should be computed as a function of a friction and cohesion term:
vn = c + μρf
y ≤ min(K
1f c ,K
2)
where
c = cohesion stress
K1 = f c coefficient for shear stress upper limit
K2 = shear stress upper limit
The coefficient c represents the cohesive strength and depends on the interface conditions. Cohesion values are reported in Table 6 alongside the μ values for the various interface conditions. Table 6 also reports the various strength limits. In addition to limiting the yield stress of the reinforcement to 60 ksi (413.7 MPa), the clamping stress f
y should be greater
than 0.05 ksi (0.34 MPa).
Discussion of the results
The performance of seven code-based models and 15 other models for interface strength were evaluated against the test data and the results obtained are summarized in Table 3. They were evaluated primarily by looking at the predicted-to-mea-sured strength ratios. Not all models were applicable to all interface and concrete types. In addition, in some cases the input parameters (such as concrete type, minimum reinforce-ment ratio, and geometrical details) necessary to employ some of the models were missing. In these cases, not applicable (n/a) was reported when summarizing the results (Table 3).
Table 3 summarizes the results (separated as a function of the interface type) in terms of mean strength ratios and associated coefficients of variation (COVs). The results are not separat-ed by concrete type, but appropriate equations are used for normalweight, sand lightweight, or all-lightweight concrete specimens.
A model that applied to a wide range of data with good accu-racy could be considered superior to a model with better accu-racy over a more limited range. Models with strict applicabil-ity restrictions could be checked against a smaller number of test results from the database. Furthermore, even in the case of poor mean predictions, models that result in a small COV had the potential to perform well after a minor modification of the parameters.
The results reported in Table 3 clearly indicate that no single design provision or strength model can be considered gener-al in nature while consistently showing high performance for all interfaces. In order to provide a single number to char-
Table 5. PCI Design Handbook friction coefficients and upper limits
Note: c = cohesion stress; K1 = fc' coefficient for shear stress upper limit; K2 = shear stress upper limit; μ = coefficient of friction. 1 ksi = 6.895 MPa.
102 PCI Journal | July–August 2020
Figure 8. Accuracy of available models and design codes. Note: CJ-R = cold joint intentionally roughened; CJ-S = cold joint not roughened, or smooth; MO-P = monolithic precracked; MO-U = monolithic uncracked; vexp = shear strength measured experimen-tally (stress); vpred = predicted shear strength (stress).
103PCI Journal | July–August 2020
acterize the model performance, a model error value em was
defined as follows:
em = (1− µi )COVii=1
4
∑The AASHTO LRFD specifications were found to give the low-est error of all models and the lowest of the code-based models. The lowest performance, with a mean v
pred /v
exp of 0.74 and a
COV of 35%, was associated with cold joint interfaces that were not roughened. This outcome was common to all models and was to be expected given the sparsity of data pertaining to cold joint interfaces that were not roughened and the uncertain-ties associated with their surface roughness conditions.
Similarly, all strength models performed poorly overall for cold joint interfaces that were not roughened. Performances similar to those achieved using the AASHTO LRFD specifications for cold joint interfaces that were not roughened could be achieved using the models of Kono et al.59 and Harries et al.25
Models that performed satisfactorily (to different extents) for interfaces other than cold joints that were not roughened were those of Lin and Chen,58 Kono et al.,59 Mattock,60 and Harries et al.25 In general, these models showed acceptable estimates of the mean strength ratios with relatively low COV values.
An alternative view of the results summarized in Table 3 is provided in Fig. 8 for the four interface types present in the database.
Accuracy of U.S. code provisions
This section presents a more detailed analysis of the perfor-mance of the interface code provisions currently in effect in the United States, namely, ACI 318, the PCI Design Hand-book, and the AASHTO LRFD specifications. In particular, a closer look was taken at how the models perform at different clamping stress ρf
y ranges for the four interface types. The
ranges were 50 to 400 psi, 401 to 800 psi, 801 to 1200 psi, and 1201to 1600 psi.
Figure 9 summarizes the results of the U.S. code accura-cy. A minimum clamping stress value greater than 50 psi (0.34 MPa) was used so that AASHTO LRFD specifications provisions would be applicable. This was done to avoid ex-cessively misleading representations of the results given that both the PCI Design Handbook and the ACI 318 shear friction equations predict zero strength for unreinforced interfaces.
With reference to Fig. 9, the following trends were observed for the individual interfaces:
• Monolithic uncracked: All three codes provided conser-vative estimates of the interface strength. Consistent with the previous section, the AASHTO LRFD specifications equations tended to provide a better overall estimate of the measured strength. The PCI Design Handbook and ACI 318 strength estimates tended to be lower than the
recorded experimental values, particularly at low clamp-ing stresses. The PCI Design Handbook predictions tended to be more accurate than those of ACI 318.
• Monolithic precracked: Both the PCI Design Handbook and ACI 318 predictions were conservative at all clamp-ing stress values, with the PCI Design Handbook strength estimates somewhat more accurate than the ACI 318 estimates. Although the AASHTO LRFD specifications are again on average more accurate than both the PCI Design Handbook and ACI 318, on occasion (for exam-ple, at low clamping stress values), the AASHTO LRFD specifications strength estimates were unconservative. This outcome had not emerged earlier when looking at the overall performance of the models. There were only four data points for clamping stress that were larger than 1600 psi (11 MPa), hence the predictions of all codes were not particularly meaningful in that range.
• Intentionally roughened cold joint: The AASHTO LRFD specifications provided the most accurate pre-dictions for intentionally roughened cold joint interfac-es. The predictions of the PCI Design Handbook and ACI 318 tended to be more conservative, with the PCI Design Handbook consistently providing more accu-rate strength estimates than ACI 318 for all clamping stress values. This suggested that both the cohesion plus friction approach of the AASHTO LRFD specifications and the effective friction approach of the PCI Design Handbook were more effective than the pure friction approach of ACI 318 at estimating the correct strength, particularly at small ρf
y values.
• Cold joint that is not roughened: ACI 318 and the PCI Design Handbook provided identical strength predictions for cold joint interfaces that were not roughened, given that the effective friction approach is not allowed for this interface type. The strength predictions were conservative at all clamping stress values, but the scatter was much larg-er than for the other interface types. The AASHTO LRFD specifications tended to be more accurate on average but provided unconservative strength estimates for low clamp-ing stresses (with high scatter). Smooth interfaces were sensitive to the interface conditions, which could have varied greatly between different experimental programs, as well as within the same experimental program. This was particularly true at low clamping stresses, when cohesion presumably played a greater role, because cohesive failure was likely to be brittle and therefore sensitive to the fine details of the loading equipment and testing procedure. In addition, fewer data points were available (and only for certain clamping stress ranges) compared with all other interface types. For these reasons, poor performance and high scatter were almost inevitable for any strength model.
Figure 10 provides another overview of the performance of the U.S. codes. The graphs show the shear strength measured experimentally (stress) v
exp as a function of the
104 PCI Journal | July–August 2020
Figure 9. Accuracy of U.S. design codes as a function of clamping stress. Note: AASHTO = American Association of State Highway and Transportation Officials’ AASHTO LRFD Bridge Design Specifications, fourth edition; ACI 318 = American Concrete Institute’s Building Code Requirements for Structural Concrete (ACI 318-14) and Commentary (ACI 318-14R); CJ-R = cold joint intentionally roughened; CJ-S = cold joint not roughened, or smooth; fy = yield stress of reinforcement; MO-P = monolithic precracked; MO-U = monolithic uncracked; PCI 7th = PCI Design Handbook: Precast and Prestressed Concrete, seventh edition; vexp = shear strength measured experimentally (stress); vpred = predicted shear strength (stress); ρ = reinforcement ratio; ρfy = clamping stress. 1 psi = 6.895 kPa.
105PCI Journal | July–August 2020
clamping stress ρfy. The results pertaining to normalweight
concrete specimens are shown on the left, and the light-weight concrete specimens are grouped on the right. The results are separated by interface type. Because current code provisions do not differentiate between monolithic un-cracked and precracked, the experimental results pertaining to these two cases are shown in the same chart and referred to simply as MO. The ACI 318, PCI Design Handbook, and AASHTO LRFD specifications strength predictions were shown alongside the experimental results. The predicted
interface shear strengths were estimated without employing any safety factors and were shown using separate lines for the minimum and maximum concrete strength f c reported in the database for that particular category.
The three codes considered provided interface strength esti-mates that differed remarkably from one another pertaining to monolithic and intentionally roughened cold joint interfaces, whereas they predicted essentially identical strengths for cold joint specimens that were not roughened.
Figure 10. Interface strength predictions of U.S. design codes. Note: Upper- and lower-bound curves correspond to the extremes of concrete strength in the relevant part of the database. AASHTO = American Association of State Highway and Transportation Officials’ AASHTO LRFD Bridge Design Specifications, fourth edition; ACI 318 = American Concrete Institute Building Code Re-quirements for Structural Concrete (ACI 318-14) and Commentary (ACI 318-14R); fy = yield stress of reinforcement; PCI 7th = PCI Design Handbook: Precast and Prestressed Concrete, seventh edition; vexp = shear strength measured experimentally (stress); ρ = reinforcement ratio; ρfy = clamping stress. 1 psi = 6.895 kPa.
106 PCI Journal | July–August 2020
The following observations were made for the normal-weight specimens:
• The AASHTO LRFD specifications provided the most accurate strength predictions for monolithic and inten-tionally roughened cold joint interfaces at all clamping stress levels. The PCI Design Handbook appeared to be more effective than ACI 318 at predicting the strength at low clamping stress values, but the 1000 psi (6.9 MPa) upper limit that is currently enforced seemed to be excessively strict and the interface strength tended to be underpredicted at medium to high clamping stress values, particularly in the case of monolithic specimens. The AASHTO LRFD specifications and ACI 318 upper limits appeared to be more consistent with the experi-mental measurements.
• The strength predictions obtained for cold joint inter-faces that were not roughened were reasonable overall, and all three codes tended to be conservative. However, there were a few specimens that failed at surprisingly low stresses (in the 250 to 500 psi [1.7 to 3.4 MPa] clamping stress range) and all provisions were uncon-servative. The scatter was very large at zero clamping stress, which made it difficult to estimate a reasonable cohesion value. In general, there were too few data points and the surface roughness conditions were too poorly defined to draw definitive conclusions about the reliability of current code provisions.
Similar observations were made for the lightweight specimens:
• The AASHTO LRFD specifications appeared to over-estimate the strength of monolithic interfaces, possibly because of an excessively large cohesion term. In con-trast, the PCI Design Handbook provided very accurate strength estimates and the ACI 318 predictions were reasonable overall, but more conservative.
• The AASHTO LRFD specifications provided the most accurate strength predictions for intentionally rough-ened cold joint interfaces. The PCI Design Handbook was somewhat more accurate than ACI 318, which tend-ed to underpredict the interface strength at low clamp-ing stress values. The AASHTO LRFD specifications upper limits also appeared to be more in line with those observed experimentally.
• All codes provided virtually identical strength predictions for cold joint interfaces that were not roughened. Although all estimates were conservative, there were too few exper-imental data points to identify clear trends and to gauge whether the codes captured those trends adequately.
Two additional observations were noted. The first, associated with practical implementation, concerned the monolithic un-cracked and monolithic precracked interfaces. The experimen-tal data clearly showed that monolithic uncracked interfaces
were stronger than monolithic precracked interfaces. Re-searchers have also been careful to differentiate between the two. However, codes do not differentiate because it is almost impossible to guarantee that a particular shear plane will remain uncracked and therefore qualify for the higher μ value that such a surface would rightfully attract. Even in a pre-stressed member, the locations where shear friction is likely to be invoked tend to be disturbed regions, such as the dapped end of a double tee, and where cracking is likely to occur. A designer would face the same dilemma of needing certainty that the potential sliding plane would remain uncracked. It therefore seems appropriate that code requirements for mono-lithic concrete should be based exclusively on the monolithic precracked data.
The second observation concerned the concrete weight reduction factor λ and concrete strength. A cursory glance at the data, for example in Fig. 4, gives the impression that lightweight concrete had lower shear friction capacity be-cause the lightweight data were low on the graph. However, three-dimensional figures such as Fig. 5 show that almost all of the lightweight specimens that have been tested have concrete strengths in the lower end of the range, but they have essentially the same shear friction strength as normal-weight specimens of the same concrete strength.
Furthermore, there was a clear trend that higher concrete strength led to higher shear friction strength, particularly under high clamping stresses. However, almost all codes penalize concrete on the basis of unit weight (lightweight versus normalweight) when that distinction is not supported by the data and fail to account for the beneficial effects of concrete strength, which is supported. Sneed28 has come to a similar conclusion about the lightweight factor. The same λ factor appears in other code provisions—for example, those pertaining to the development length of reinforce-ment. Although that behavior lies outside the scope of this study, it would be of interest to determine whether the λ factor is justified. It is worth noting that, in ACI 318, devel-opment lengths of bar reinforcement do depend on concrete strength, while those for strands do not.
Conclusion
This paper presents a database of 509 interface shear test results collected from the literature. The database was analyzed critically to identify major trends and current research gaps. The data collected were also used to gauge the reliability of six code provisions and 14 strength mod-els. The findings from this research support the following main conclusions:
• The interface shear strength was mostly (but to differ-ent extents) affected by the interface conditions (that is, roughness), the clamping stress, and the concrete strength. The roughness level and the clamping stress played an important role for all interfaces, whereas the concrete strength mostly affected the interface response
107PCI Journal | July–August 2020
of monolithic uncracked and intentionally roughened cold joint interfaces.
• A cohesion component existed in all interfaces, but the ex-perimental data analyzed suggested that it could be signif-icant for intentionally roughened cold joint and monolithic uncracked interfaces, while it could be neglected (unless supported by new experimental evidence) for monolithic precracked and not roughened cold joint interfaces.
• The data did not support a dependence of shear strength on steel yield strength for a given clamping stress. How-ever, the data were sparse for f
y > 100 ksi (689.5 MPa)
and more research is needed in this area.
• The interface shear strength of lightweight concrete inter-faces was, on average, no less than that of normalweight concrete interfaces for the same concrete strength. The widely used λ factor appeared not to be justified.
• The compressive strength of the concrete played a role, but primarily for high ρf
y values. In that region, the data were
too sparse to permit development of a suitable relationship.
Of all the design codes considered, the AASHTO LRFD specifications provided the most accurate strength estimates across all interfaces. The PCI Design Handbook was general-ly more accurate than ACI 318, which tended to consistently underestimate the interface strength for all interface types and at all clamping stress levels. Among the researchers’ strength models analyzed, those of Lin and Chen,58 Kono et al.,59 Mat-tock,60 and Harries et al.25 were the most accurate overall, but they were still less accurate overall than the AASHTO LRFD specifications. In general, all codes and models fell short when attempting to predict the strength of not roughened cold joint interfaces. This was due to the sparsity of data and to the poorly defined roughness conditions associated with this in-terface type, which therefore led to a wide variety of different surface conditions in tests.
Recommendations
• Based on the experimental observations, it is recommend-ed that the current limit on the reinforcement yield stress be raised from 60 to 80 ksi (413.7 to 551.6 MPa). Other interface strength upper limits should also be revised. In particular, the PCI Design Handbook upper limits could be raised to be consistent with the AASHTO LRFD spec-ifications and ACI 318 upper bounds.
• In the current shear friction code provisions, lightweight concrete usage is permitted with the use of concrete modification factor λ. However, the experimental results analyzed suggest that there is no inherent weakness asso-ciated with using lightweight concrete. It is recommend-ed that the concrete modification factor be removed from the equations. This recommendation is consistent with that of Sneed et al.28
• If reliable models for cold-joint interfaces are desired, there is a need for more experiments that better define interface roughness or a roughening procedure. Such a definition should enter the design practice to guarantee that certain standards and consistency are achieved. More specifically, even if current ACI 318 standards are not met, a common definition of roughness should still be available to avoid the implementation of in-terfaces with drastically different roughness and, in turn, strength.
• Based on the analysis of the available experimental data and on the observation of the best strength predictions obtained using design codes (that is, the AASHTO LRFD specifications) and strength models (such as Mattock60 and Harries et al.25) that contain both a cohesion and a friction term, it is recommended that this form be adopted universally in the future.
• It is further recommended that the cohesion term in the model be a function of both the interface conditions and the concrete strength (as suggested by both Mattock60 and Harries et al.25) rather than considering only the interface conditions (as currently done in the AASHTO LRFD specifications).
Acknowledgments
This research was conducted with the sponsorship of PCI. The authors wish to thank Roger Becker of PCI, Harry Gleich of Metromont Precast, Ned Cleland of Blue Ridge Design Inc., and Neil Hawkins, who served as advisors to this project. Their assistance and input are greatly appreciated.
References
1. Birkeland, P. W., and H. W. Birkeland, 1966. “Connec-tions in Precast Concrete Construction.” Journal of the American Concrete Institute 63 (3): 345–368.
2. Hanson, N. W. 1960. “Precast-Prestressed Concrete Bridges: (2), Horizontal Shear Connections.” PCA Re-search and Development Laboratories 2 (2): 38–58.
3. Saemann, J. C., and G. W. Washa. 1964. “Horizontal Shear Connections between Precast Beams and Cast-in-Place.” ACI Journal Proceedings 61 (11): 1383–1410.
4. Ivey, D. L., and E. Buth. 1967. “Shear Capacity of Light-weight Concrete Beams.” ACI Journal 64 (10): 634–643.
5. Hofbeck, J. A., I. O. Ibrahim, and A. H. Mattock. 1969. “Shear Transfer in Reinforced Concrete.” ACI Journal Proceedings 66 (2): 119–128.
6. Vangsirirungruang, K. 1971. “Effect of Normal Compres-sive Stresses on Shear Transfer in Reinforced Concrete.” MS thesis, University of Washington, Seattle.
108 PCI Journal | July–August 2020
7. Chatterjee, P. 1971. “Shear Transfer in Reinforced Con-crete.” MS thesis, University of Washington, Seattle.
8. Dulacska, H. 1972. “Dowel Action of Reinforcement Crossing Cracks in Concrete.” ACI Journal Proceedings 69 (12): 754–757.
9. Mattock, A. H., and N. M. Hawkins. 1972. “Shear Transfer in Reinforced Concrete—Recent Research.” PCI Journal 17 (2): 55–75.
10. Paulay, T., R. Park, and M. H. Phillips. 1974. “Hori-zontal Construction Joints in Cast-in-Place Reinforced Concrete.” In ACI Special Publication SP-42, 599–616. Farmington Hills, MI: ACI.
11. Mattock, A. H., L. Johal, and H. C. Chow. 1975. “Shear Transfer in Reinforced Concrete with Moment or Tension Acting across the Shear Plane.” PCI Journal 20 (4): 76–93.
12. Mattock, A. H. 1976. “Shear Transfer under Monotonic Loading across an Interface between Concretes Cast at Different Times.” Department of Civil Engineering report SM 76-3. Seattle, WA: University of Washington.
13. Mattock, A. H., W. K. Li, and T. C. Wang. 1976. “Shear Transfer in Lightweight Reinforced Concrete.” PCI Jour-nal 21 (1): 20–39.
14. Walraven, J. C., and H. W. Reinhardt. 1981. “Theory and Experiments on the Mechanical Behaviour of Cracks in Plain and Reinforced Concrete Subjected to Shear Load-ing.” HERON 26 (1A): 1–68.
15. Bass, R. A., R. L. Carrasquillo, and J. O. Jirsa. 1989. “Shear Transfer across New and Existing Concrete Inter-faces.” ACI Structural Journal 86 (4): 383–393.
16. Hoff, G. C. 1993. “High Strength Lightweight Aggregate Concrete for Arctic Applications—Part 3: Structural Pa-rameters.” In ACI Special Publication SP-136, 175–246. Farmington Hills, MI: ACI.
17. Loov, R. E., and A. K. Patnaik. 1994. “Horizontal Shear Strength of Composite Concrete Beams with a Rough Interface.” PCI Journal 39 (1): 48–69.
18. Valluvan, R., M. E. Kreger, and J. O. Jirsa. 1999. “Eval-uation of ACI 318-95 Shear-Friction Provisions.” ACI Structural Journal 96 (4): 473–481.
19. Gohnert, M. 2000. “Proposed Theory to Determine the Hor-izontal Shear Between Composite Precast and In Situ Con-crete.” Cement and Concrete Composites 22 (6): 469–476.
20. Kahn, L. F., and A. D. Mitchell. 2002. “Shear Friction Tests with High-Strength Concrete.” ACI Structural Jour-nal 99 (1): 98–103.
21. Papanicolaou, C. G., and T. C. Triantafillou. 2002. “Shear Transfer Capacity along Pumice Aggregate Concrete and High-Performance Concrete Interfaces.” Materials and Structures 35 (4): 237–245.
22. Gohnert, M. 2003. “Horizontal Shear Transfer across a Roughened Surface.” Cement and Concrete Composites 25 (3): 379–385.
23. Hawkins, N. M., and D. A. Kuchma. 2007. “Application of LFRD Bridge Design Specifications to High Strength Structural Concrete: Shear Provisions.” National Coop-erative Highway Research Program report 579, National Academies of Sciences, Engineering, and Medicine. Washington, DC: The National Academies Press. https://doi.org/10.17226/17616.
24. Mansur, M. A., T. Vinayagam, and K. H. Tan. 2008. “Shear Transfer across a Crack in Reinforced High-Strength Concrete.” Journal of Materials in Civil Engi-neering 20 (4): 294–302.
25. Harries, K. A., G. Zeno, and B. Shahrooz. 2012. “Toward an Improved Understanding of Shear-Friction Behavior.” ACI Structural Journal 109 (6): 835–844.
26. Shaw, D., and L. H. Sneed. 2014. “Interface Shear Trans-fer of Lightweight-Aggregate Concretes Cast at Different Times.” PCI Journal 59 (3): 130–144.
27. Echegaray-Oviedo, J., E. Cuenca, J. Navarro-Gregori, and P. Serna. 2014. “Influence of the Fiber Reinforce-ment in Concrete under Direct Shear.” In 10th fib Inter-national PhD Symposium in Civil Engineering Proceed-ings, 415–424. Quebec City, QC, Canada: Université Laval.
28. Sneed, L. H., K. Krc, S. Wermager, and D. Meinheit. 2016. “Interface Shear Transfer of Lightweight-Aggre-gate Concretes.” PCI Journal 61 (2): 38–55.
29. PCI Industry Handbook Committee. 2004. PCI Design Handbook: Precast and Prestressed Concrete Institute. MNL-120. 6th ed. Chicago, IL: PCI.
30. PCI Industry Handbook Committee. 2010. PCI Design Handbook: Precast and Prestressed Concrete Institute. MNL-120. 7th ed. Chicago, IL: PCI.
31. AASHTO (American Association of State Highway and Transportation Officials). 2007. AASHTO LRFD Bridge Design Specifications. 4th ed., SI units. Washington, DC: AASHTO.
32. ACI (American Concrete Institute) Committee 318. 2014. Building Code Requirements for Structural Concrete (ACI 318-14) and Commentary (ACI 318R-14). Farming-ton Hills, MI: ACI.
109PCI Journal | July–August 2020
33. Tanner, J. A. 2008. “Calculating Shear-Friction Using Effective Coefficient of Friction.” PCI Journal 53 (3): 114–120.
34. Nagle, T. J., and D. A. Kuchma 2007. “Nontraditional Limitations on the Shear Capacity of Prestressed Con-crete Girders.” Newmark Structural Engineering Labora-tory report 003, University of Illinois at Urbana-Cham-paign.
35. Mattock, A. H. 1974. “Shear Transfer in Concrete Having Reinforcement at an Angle to the Shear Plane.” In ACI Special Publication SP-42, 17–42. Farmington Hills, MI: ACI.
36. Crane, C. K. 2010. “Shear and Shear Friction of Ul-tra-High Performance Concrete Bridge Girders.” PhD diss., Georgia Institute of Technology.
37. Frenay, J. W. 1985. “Shear Transfer across a Single Crack in Reinforced Concrete under Sustained Loading. Part I: Experiments.” Report 5-85-5, Stevin Laboratory, Concrete Structures. Delft University of Technology, Faculty Civil Engineering and Geosciences. http:// resolver.tudelft.nl/uuid:6cd80da5-245b-4bd5-b21a -c48d37f57b6e.
38. Calvi, P. M., E. C. Bentz, and M. P. Collins. 2016. “Reversed Cyclic Experiments on Shear Stress Transfer across Cracks in Reinforced Concrete Elements.” ACI Structural Journal 113 (4): 851–859.
39. Calvi, P. M., E. C. Bentz, and M. P. Collins. 2017. “Pure Mechanics Crack Model for Shear Stress Transfer in Cracked Reinforced Concrete.” ACI Structural Journal 114 (2): 545–554.
40. Davaadorj, O. 2018. “Shear Stress Transfer across Con-crete-to-Concrete and Steel-to-Concrete Interfaces.” MS thesis, Department of Civil and Environmental Engineer-ing, University of Washington.
41. Barbosa, A. R, D. Trejo, and D. Nielson. 2017. “Effect of High-Strength Reinforcement Steel on Shear Friction Be-havior.” Journal of Bridge Engineering 22 (8): 04017038.
42. CSA (Canadian Standards Association). 2014. Design of Concrete Structures. A23.3-04. Toronto, ON, Canada: CSA.
43. CEB (Comité Euro-International du Béton). 1990. “Mod-el Code for Concrete Structures.” In CEB-FIP Model Code 1990. Lausanne, Switzerland: CEB.
44. CEN (Comité Européen de Normalisation). 2004. “De-sign of Concrete Structures.” In Eurocode 2, Part 1.1: General Rules and Rules for Buildings [with corrigendum of 16 January 2008]. Brussels, Belgium: CEN.
45. Concrete Technology Associates. 1974. “Composite Systems without Roughness.” Technical bulletin 74-B6, CTA 35, 271-316.
46. Concrete Technology Associates. 1976. “Composite Systems without Ties.” Technical bulletin 76-B4, CTA 36, 317-360.
47. Walraven, J., and J. Stroband. 1994. “Shear-Friction in High-Strength Concrete” In ACI Special Publication SP-42, 311–330. Farmington Hills, MI: ACI.
48. Pincheira, J. A., M. Oliva, and F. I. Kusumo-Rahardjo. 1998. “Tests on Double Tee Flange Connectors Subjected to Monotonic and Cyclic Loading.” PCI Journal 43 (3): 82–96.
49. Scott, J. 2010. “Interface Shear Strength in Lightweight Concrete Bridge Girders.” MS thesis, Virginia Polytech-nic Institute, Blacksburg.
50. Sagaseta, J., and R. L. Vollum. 2011. “Influence of Aggregate Fracture on Shear Transfer through Cracks in Reinforced Concrete.” Magazine of Concrete Research 63 (2): 119–137.
51. Rahal, K. N., and A. L. Al-Khaleefi. 2015. “Shear-Fric-tion Behavior of Recycled and Natural Aggregate Con-crete—An Experimental Investigation.” ACI Structural Journal 112 (6): 725–733.
52. Mast, R. F. 1968. “Auxiliary Reinforcement in Concrete Connections.” ASCE Journal of the Structural Division Proceedings 94 (ST6): 1485–1504.
53. Raths, C. H. 1977. “Reader Comments: Design Proposals for Reinforced Concrete Corbels.” PCI Journal 21 (3): 93–98.
54. Loov, R. E. 1978. “Design of Precast Connections.” Paper presented at a seminar organized by Compa International Pte. Ltd. Singapore, Sep. 25–27, 1978.
55. Shaikh, F. A. 1978. “Proposed Revisions to Shear-Fric-tion Provisions.” PCI Journal 23 (2): 12–21.
56. Hsu, T. T. C., S. T. Mau, and B. Chen. 1987. “Theory of Shear Transfer Strength of Reinforced Concrete.” ACI Structural Journal 84 (2): 149–160.
57. Tsoukantas, S. G., and T. P. Tassios. 1989. “Shear Resistance of Connections between Reinforced Concrete Linear Precast Elements.” ACI Structural Journal 86 (3): 242–249.
58. Lin, I. J., and Y. L. Chen. 1989. “Shear Transfer across a Crack in Reinforced High Strength Concrete.” In Pro-ceedings of the 2nd East Asia–Pacific Conference on Structural Engineering and Construction, Chiang Mai, Thailand. 505–510, January 1989.
110 PCI Journal | July–August 2020
59. Kono, S., H. Tanaka, and F. Watanabe. 2003. “Interface Shear Transfer for High Strength Concrete and High Strength Shear Friction Reinforcement.” In High Per-formance Materials in Bridges, ed. A. Azizinamini, A. Yakel, and M. Abdelrahman, 319–328. Reston, VA: American Society of Civil Engineers.
60. Mattock, A. H. 2001. “Shear-Friction and High-Strength Concrete.” ACI Structural Journal 98 (1): 50–59.
61. Walraven, J., J. Frénay, and A. Pruijssers. (1987). “Influ-ence of Concrete Strength and Load History on the Shear Friction Capacity of Concrete Members.” PCI Journal, 32 (1): 66–84.
Notation
ag = maximum aggregate size
Ac = area of shear interface
Acr = shear interface area
c = cohesion stress
ds = diameter of steel reinforcement
em = model error
f c = concrete compressive strength
fy = yield stress of reinforcement
K1 = f c coefficient for shear stress upper limit
K2 = shear stress upper limit
L = beam length
P = applied load
vexp
= shear strength measured experimentally (stress)
vn = nominal shear strength (stress)
vpred
= predicted shear strength (stress)
Vexp
= shear strength measured experimentally (force)
Vn,max
= nominal shear strength (force)
Vpred
= predicted shear strength (force)
Vu = shear demand
λ = concrete weight reduction factor
μ = interface friction coefficient
μe = effective coefficient of friction
ρ = reinforcement ratio
ρfy = clamping stress
φ = strength reduction factor
111PCI Journal | July–August 2020
About the authors
Otgonchimeg Davaadorj is a former master’s student in structural engineering in the Department of Civil and Environ-mental Engineering at the Univer-sity of Washington, Seattle.
Paolo M. Calvi, PhD, is an assistant professor in the Depart-ment of Civil and Environmental Engineering at the University of Washington, Seattle.
John F. Stanton, PhD, is a profes-sor in the Department of Civil and Environmental Engineering at the University of Washington, Seattle.
Abstract
This paper presents results of a database of interface shear experiments collected from the literature and analyzed to identify the main parameters affecting the strength of concrete-to-concrete interfaces subjected to shear loads and to gauge the reliability of major international codes and interface strength models. The database included 509 push-off and pull-off specimens, with steel reinforcement normal to the interface and subjected to monotonic pure shear loading. The experi-mental data were analyzed mainly in terms of interface type, clamping stress, concrete compressive strength, concrete unit weight, and steel strength.
The analysis of the database revealed that clamping stress and interface type were the main parameters influencing the interface strength but that the con-crete strength could play an important role as well. In contrast, it was found that the concrete unit weight appeared to have no effect on the interface strength. Of the code provisions considered, it was found that the American Association of State Highway and Transpor-tation Officials’ AASHTO LRFD Bridge Design Spec-ifications shear friction equations were the most accu-rate at predicting the interface strength for all interface types and conditions, while the PCI Design Handbook: Precast and Prestressed Concrete and American Concrete Institute’s Building Code Requirements for Structural Concrete (ACI 318-14) and Commentary (ACI 318R-14) equations tended to underestimate the strength, with the PCI Design Handbook being gener-ally more accurate than ACI 318-14.
This paper was reviewed in accordance with the Precast/Prestressed Concrete Institute’s peer-review process.
Reader comments
Please address any reader comments to PCI Journal editor-in-chief Tom Klemens at [email protected] or Precast/Prestressed Concrete Institute, c/o PCI Jour-nal, 8770 W. Bryn Mawr Ave., Suite 1150, Chicago, IL 60631. J
![Asymptotic behavior of singularly perturbed control …€¦ · Asymptotic behavior of singularly perturbed control ... [Lions, Papanicolau, Varadhan 1986]; ... Asymptotic behavior](https://static.documents.pub/doc/80x56/5b7c19bc7f8b9a9d078b9b98/asymptotic-behavior-of-singularly-perturbed-control-asymptotic-behavior-of-singularly.jpg)
![Homogenization and asymptotics for small transaction costs ...touzi/Possamai-Soner-Touzi-23.12.2012.pdf · of Papanicolau & Varadhan [33] and of Souganidis [43]. However, ... to Lions](https://static.documents.pub/doc/80x56/5a9925f37f8b9ab6188d2f01/homogenization-and-asymptotics-for-small-transaction-costs-touzipossamai-soner-touzi-23122012pdfof.jpg)
